Low copy number plasmids in bacteria are of interest for two principle reasons. First, they act in many ways like small, dispensable chromosomes within the cell, and are therefore tractable models for the study of chromosome replication and segregation. Second, they are of considerable medical importance. They are the transmissible elements that spread antibiotic resistance among pathogenic bacteria and in some cases, are the determinants of the virulence of bacterial infection in human infectious disease. The spread of antibiotic resistance threatens to make antibiotic therapy virtually useless in the next few decades. In addition, pathogenic bacteria containing virulence plasmids are increasingly the ultimate cause of death of cancer patients whose immune systems are often compromised by disease progression or chemotherapy. It is therefore of importance to try to understand how these plasmids are stably maintained in the bacterial population as a first step toward developing novel strategies for infectious disease therapy and remediation of plasmid spread. We are particularly interested in the mechanisms that plasmids use to ensure their proper segregation to daughter cells. We study a family of elements known as partition genes (the P1par family), that are responsible for the segregation of several types of plasmid including the virulence plasmids of Salmonella and Shigella species responsible for enteric disease, and of Yersinia pestis;the causative organism for bubonic plague. In each case, we have shown that segregation is achieved by recognition of a cis-acting site parS, analogous to a centromere, and two plasmid encoded proteins, ParA and ParB.. ParB binds specifically to parS and ParA is an ATPase that may be a motor for moving the plasmid during segregation. Members of the P1par family show unique species specificities. This is important, because, otherwise, plasmids of different types would compete with each other, limiting their spread in nature. We have discovered that these species specificities reside in a novel interaction between the ParB protein and the parS site. This is not the interaction that provides the energy for ParB binding to the site. Rather, it is a special contact between the ParB N-terminus and a short motif in parS termed the B box. By changing the B box sequence by as little as one base, we can change the specificity of the system from one species to another. This mechanism appears to be a novel type of DNA-protein recognition that may have broad implications for how proteins act at a specific site when other potential binding sites exist. We have further explored the hypothesis that the contact between the BoxB sequences in the cis-acting partition site and the ParB protein are responsible for the species specificity of members of the P1par family of partition elements. Using the recently published crystal structure of the P1 ParB protein as a guide, we have been able to pinpoint the position on the ParB protein surface that specifies recognition of species determinants on the individual plasmid DNA parS sites. By changing a single amino acid in the ParB protein we have been able to switch its recognition specificity completely to that of a different species. We have purified these altered proteins and have been able to show that thier altered specificity does not reside in a change in DNA binding ability. Rather, it appears that the BoxB DNA site acts as a specific activator of the partition complex for plasmid segregation. This has important implications for models of partition. We are currently testing these conclusions by studying the segregation of the plasmid DNA by fluorescence photomicroscopy in living cells. In addition, we have carried out a collaboration with Dr. Xinhua Ji of the Biomolecular Structure Section (Chystallography Laboratory, CCR), to model the molecular structures of the altered interactions. The results fit well with the genetic observations and confirm that a single DNA base-amino acid side chain interaction is key to determining the specificity of these systems. Our development of automated data collection and analysis of fluorescent foci in living bacterial cells have given us an opportunity to re-visit the behavior of the P1 plasmid in E. coli. The results have been surprising, and suggest that much of the current literature on this subject is wrong. It has been claimed that P1, and related plasmid types, replicate in the cell center and segregate to the cell quarter positions, ready to be at the center of the daughter cells at cell division. Thus, the emphasis has been on searching for host sites involved in segregation. Now, we find that there is no clear preference for fixed positions within the cell. The mean positions of multiple plasmid foci are evenly spaced from each other and from the cell poles. The dynamics of the segregation process in time-lapse and time-stamp microscopy reveals a considerable amount of movement of plasmid foci both away from and toward each other. Thus, the population and time-averaged distributions are even, but individual cell patterns vary considerably. replication and segregation can occur from almost any position in the cell. Moreover, similar numbers of focus splitting and pairing events are observed in the population. This suggests a model for partition in which plasmids that are paired either as a result of replication or by fairly frequent chance encounters. They then move rapidly away from each other along the cell axis, maintaining a dynamic distribution of copies approximating an even distribution along the long cell axis. These observations have important implications for the mechanism for physical separation of the plasmids. This year, we have developed a mathematical simulation of plasmid segregation that resembles the type of model known as a Monte Carlo simulation. The simulation is derived from the rules for segregation that we have p[roposed to account for our experimental observations. The simulation produces plasmid segregation fidelity and spacial distributions of plasmid copies within the cell that closely resemble the experimental data in all details. The simulation greatly increases our confidence that the new picture of plasmid segregation that is emerging from our studies is correct. We have developed a system by which the proteins required for P1 segregation can be depleted from the cell in response to a temperature shift. The results provide strong support for our model for segregation. The pseudo-even distribution of plasmid copies breaks down progressively with time when the partition proteins are removed. Spacing becomes gradually random, with a progressive increase in cells containing all plasmid copies on one side of the cell. The ultimate destination for the plasmid copies appears to be accumulation in the cell oples and subsequent loss from one daughter cell at each division.